Dr. William Brady on Social Media, Moral Outrage and Polarization

What It Covers

Dr. William Brady explains how social media algorithms amplify moral outrage through social learning mechanisms, creating feedback loops where users learn to express more outrage for likes and shares, while bots and extreme minorities dominate political discourse.

Key Questions Answered

• •Prime Information Bias: Humans naturally prioritize prestigious, in-group, moralized, and emotional content for efficient social learning. Social media algorithms exploit this by amplifying such content to maximize engagement and advertising revenue, creating oversaturation that distorts perception of social norms and makes extreme views appear more common than they actually are.
• •Outrage Amplification Loop: When users receive likes and shares for expressing moral outrage, they become conditioned to post more outrage content in the future. This variable reinforcement pattern mimics psychological conditioning, with algorithms further amplifying outraged posts, creating a feedback loop where users incorrectly infer that everyone approves of outrage-driven communication.
• •Ideological Distribution Effects: Research shows 75 percent of polarizing content comes from just 20 percent of users on the political extremes. Moderate users in the middle 60 percent of the distribution are more sensitive to social feedback and vulnerable to learning outrage behaviors, while extreme users express outrage habitually regardless of likes received.
• •Representative Diversification Solution: Algorithms can be redesigned to show a more accurate distribution of political views rather than amplifying extreme voices. Testing on BlueSky platform demonstrates this approach maintains user engagement while reducing exhaustion from toxic content, as surveys show most users dislike the negative moral content they currently see but click anyway.
• •Foreign Interference Patterns: Analysis of Internet Research Agency troll accounts from 2016 reveals they systematically post content containing highly polarizing language and misinformation designed to elicit maximum moral outrage. This misinformation produces significantly more outrage than legitimate news sources, exploiting algorithmic amplification to divide populations and reduce trust in information ecosystems.